Share Your Research, Maximize Your Social Impacts
Main Menu
Searching By
Main Menu
PARTNERS
PhD in Machine Learning for Molecular Sciences at FU Berlin
Country/Region : Germany
Website : http://fu-berlin.de
Description
The research group of Frank Noé at FU Berlin is highly interdisciplinary, with members from physics, engineering, mathematics, computer science and chemistry. Our aim is to develop fundamental computational methods to solve hard problems in the natural sciences, in particular the molecular sciences (biophysics and theoretical chemistry).
We aim at disseminating our methods in high-quality open source software. We conduct research is conducted in teams, each team has a bandwidth of theoretical, applied and software-oriented researchers and works on an overarching goal.
We are currently looking for PhD applicants especially in the following two areas:
1) Molecular and material design with generative models and development of generative models (variational autoencoders, generative adversarial nets and beyond) for complex structures with physical constraints.
2) Machine learning of dynamical models from time-series data [1-3]. Learning dynamical equations, automatic model reduction and coarse-graining and development of open-source software.
Applicants should have a solid mathematical education and ideally have significant experience and interest in software development. Prior knowledge in machine learning and the typical frameworks (e.g. Tensorflow, Theano, PyTorch etc.) is a plus.
Please send CV, 1-2 reference contacts, BSc+MSc transcript or course/grade list and a statement of interest to
frank.noe-AT-fu-berlin.de
Some recent work:
[1] Mardt, Pasquali, Wu and Noé: VAMPnets: deep learning of molecular kinetics
https://arxiv.org/abs/1710.06012, Nature Communications (in press)
[2] PyEMMA Software: www.pyemma.org. See also M Scherer et al: PyEMMA 2: A Software Package for Estimation, Validation, and Analysis of Markov Models. J. Chem. Theory Comput., 11 . pp. 5525-5542.
[3] Plattner, Doerr, De Fabritiis and Noé: Complete protein-protein association kinetics in atomic detail
revealed by molecular dynamics simulation and Markov modelling. Nature Chemistry 9, 1005 (2017)
We aim at disseminating our methods in high-quality open source software. We conduct research is conducted in teams, each team has a bandwidth of theoretical, applied and software-oriented researchers and works on an overarching goal.
We are currently looking for PhD applicants especially in the following two areas:
1) Molecular and material design with generative models and development of generative models (variational autoencoders, generative adversarial nets and beyond) for complex structures with physical constraints.
2) Machine learning of dynamical models from time-series data [1-3]. Learning dynamical equations, automatic model reduction and coarse-graining and development of open-source software.
Applicants should have a solid mathematical education and ideally have significant experience and interest in software development. Prior knowledge in machine learning and the typical frameworks (e.g. Tensorflow, Theano, PyTorch etc.) is a plus.
Please send CV, 1-2 reference contacts, BSc+MSc transcript or course/grade list and a statement of interest to
frank.noe-AT-fu-berlin.de
Some recent work:
[1] Mardt, Pasquali, Wu and Noé: VAMPnets: deep learning of molecular kinetics
https://arxiv.org/abs/1710.06012, Nature Communications (in press)
[2] PyEMMA Software: www.pyemma.org. See also M Scherer et al: PyEMMA 2: A Software Package for Estimation, Validation, and Analysis of Markov Models. J. Chem. Theory Comput., 11 . pp. 5525-5542.
[3] Plattner, Doerr, De Fabritiis and Noé: Complete protein-protein association kinetics in atomic detail
revealed by molecular dynamics simulation and Markov modelling. Nature Chemistry 9, 1005 (2017)
Last modified: 2017-12-04 13:18:01